Variational Hybrid Monte Carlo for Efficient Multi-Modal Data Sampling

Variational Hybrid Monte Carlo for Efficient Multi-Modal Data Sampling

The Hamiltonian Monte Carlo (HMC) sampling algorithm exploits Hamiltonian dynamics to construct efficient Markov Chain Monte Carlo (MCMC), which has become increasingly popular in machine learning and statistics. Since HMC uses the gradient information of the target distribution, it can explore the...

Full description

Bibliographic Details
Main Authors: Shiliang Sun, Jing Zhao, Minghao Gu, Shanhu Wang
Format: Article
Language:English
Published: MDPI AG 2023-03-01
Series:Entropy
Subjects:
Markov chain Monte Carlo
Hamiltonian Monte Carlo
Langevin dynamics
multi-modal sampling
variational distribution
Online Access:https://www.mdpi.com/1099-4300/25/4/560

Internet

https://www.mdpi.com/1099-4300/25/4/560

Similar Items